Self-Organizing Maps as a New Tool for Classification of Plants at Lower Hierarchical Levels

Abstract

A chemotaxonomic analysis is described of a database containing various types of compounds from the Heliantheae tribe (Asteraceae) using Self-Organizing Maps (SOM). The numbers of occurrences of 9 chemical classes in different taxa of the tribe were used as variables. The study shows that SOM applied to chemical data can contribute to differentiate genera, subtribes, and groups of subtribes (subtribe branches), as well as to tribal and subtribal classifications of Heliantheae, exhibiting a high hit percentage comparable to that of an expert performance, and in agreement with the previous tribe classification proposed by Stuessy.